On Thursday, March 15, 2012 11:26:00 PM Paolo Bonzini wrote:
> Il 15/03/2012 20:14, Michael Roth ha scritto:
> > Unfortunately the BSOD is too early to get a kernel dump, but here are
> > the error params in case that's useful:
> >
> > 0xFC0000096, 0xF800027F6200, 0x0, 0x0
> >
> > Let me know if you need any other details.
>
> You can use the Windows Boot Debugger. (Disclaimer: I only used it
> once). Enable it with "bcdedit /bootdebug {current} ON".
>
It is not a loader problem. You even don't need to turn halbreakpoint option
on.
> To use it, you need two Windows VMs (the one that fails and one with
> WinDbg installed). It is the same as doing normal kernel debugging
> except that the debuggee will hang waiting for a connection even before
> showing the boot loader prompt. You need to connect the serial consoles
> of the VMs; I usually use socat like
>
> socat "PIPE:$tty1,unlink-close=0,noctty=1" \
> "PIPE:$tty2,unlink-close=0,noctty=1
>
> with the tty names taken from libvirt's XML. It probably works with
> Unix sockets too but I never tried.
>
> From WinDbg you should be able to get a backtrace. If you never used
> WinDbg, remember to set it up for downloading symbols:
> http://support.microsoft.com/kb/311503
>
> HTH,
>
> Paolo